CN110222844A - A kind of compressor performance prediction technique based on artificial neural network - Google Patents
A kind of compressor performance prediction technique based on artificial neural network Download PDFInfo
- Publication number
- CN110222844A CN110222844A CN201910464786.5A CN201910464786A CN110222844A CN 110222844 A CN110222844 A CN 110222844A CN 201910464786 A CN201910464786 A CN 201910464786A CN 110222844 A CN110222844 A CN 110222844A
- Authority
- CN
- China
- Prior art keywords
- layer
- compressor performance
- node
- weight
- output
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N5/00—Computing arrangements using knowledge-based models
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Software Systems (AREA)
- Data Mining & Analysis (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Computational Linguistics (AREA)
- Mathematical Physics (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
A kind of compressor performance prediction technique based on artificial neural network provided by the invention, the following steps are included: acquisition compressor performance data, and normalization process is carried out, it is used as training sample and test sample later, wherein, compressor performance data include flow, pressure ratio and revolving speed；Artificial nerve network model training is carried out to the training sample using BP algorithm, obtains trained artificial nerve network model；Test sample is inputted the artificial nerve network model that above-mentioned training obtains to detect compressor performance；Prediction technique of the invention improves the precision of prediction of the compressor performance far from sample number strong point, while reducing forecast cost.
Description
Technical field
The invention belongs to compressor performance electric powder prediction, in particular to a kind of compressor based on artificial neural network
Performance prediction method.
Background technique
The off design performance of centrifugal compressor plays critically important effect in its design, experiment and operation.And performance is bent
Line is exactly a class ourve for portraying centrifugal compressor off design performance；The main path for obtaining performance curve at present is experiment measurement
Or performance conversion is carried out based on data with existing.The former is the direct method that a kind of pair of machine performance is surveyed, and is obtained
Curve is although true and reliable, but experimental expenses is expensive.Latter solution is the data with existing or practical machine according to molding machine
The a small amount of experimental data of device converts to performance under certain similarity hypothesis.Though this method can reduce experiment,
The degree that it is too dependent on the accuracy of similarity hypothesis and is satisfied.Under normal conditions, the resulting result that converts can only protect
The Accurate Prediction near given data is demonstrate,proved, the prediction far from the data point is then often larger with actual difference, or even is not consistent very much.
Therefore, the not only economic but also complete performance curve that reliably obtains is that centrifugal compressor research and projector pursue for a long time
Target；Therefore, the compressor performance precision of prediction using the prediction technique of advanced low cost and raising far from data point is very
It is necessary.
Summary of the invention
The compressor performance prediction technique based on artificial neural network that the purpose of the present invention is to provide a kind of solves existing
The defect of high, far from data point the prediction result inaccuracy of the procurement cost of some centrifugal compressor performance curves.
In order to achieve the above object, the technical solution adopted by the present invention is that:
A kind of compressor performance prediction technique based on artificial neural network provided by the invention, comprising the following steps:
Compressor performance data are acquired, and carry out normalization process, are used as training sample and test sample later, wherein
Compressor performance data include flow, pressure ratio and revolving speed；
Artificial nerve network model training is carried out to the training sample using BP algorithm, obtains trained artificial neuron
Network model；
Test sample is inputted the artificial nerve network model that above-mentioned training obtains to detect compressor performance.
Preferably, normalization process is carried out to collected compressor performance data, specifically included:
By collected compressor performance data by regulator, it is mapped to (0, l) section, wherein described regular
Change function are as follows:
Wherein, x is compressor performance data before normalization；Y is compressor performance data after normalization, and n is compressor
The number of energy data；For the average value of x；xmaxFor the maximum value of x before normalization.
Preferably, artificial nerve network model training is carried out using BP algorithm to institute to the training sample using BP algorithm
It states training sample and carries out artificial nerve network model training, specifically include:
S1 carries out forward-propagating to training sample, and calculates the error between the real output value and desired value of each node
Value；
S2, judges whether error amount obtained in S1 meets index request, if not meeting, carries out back-propagation process,
Into S3；If meeting, artificial nerve network model training is completed, is terminated；
S3 carries out back-propagation process using gradient descent method, obtains output layer right value update using compound derivation later
Right value update formula between formula and input layer and hidden layer updates output layer weight, defeated according to above-mentioned more new formula respectively
Enter the weight between layer and hidden layer；
S4, it is into S1, the weight between updated output layer weight, input layer and hidden layer obtained in S3 is defeated
Enter the input layer in S1 in forward-propagating.
Preferably, in S1, during forward-propagating, the output of hidden layer node are as follows:
Export the output of node layer are as follows:
Wherein, f (...) is activation primitive, and mathematic(al) representation is f (x)=1/ (1+e-x)；xnFor n-th of input node
Value, wih,knIndicate the weight between input layer and hidden layer, who,jiIndicate the weight between hidden layer and output layer, niIndicate defeated
Enter the number of nodes of layer, nhFor the number of nodes of hidden layer, noIndicate the number of nodes of output layer；bkFor the biasing of k-th of node；bjFor
The biasing of j-th of node.
Preferably, in S1, during forward-propagating, the error between the real output value and desired value of each node is solved
The function of value are as follows:
Wherein, m is training sample number；dkIt (i) is desired output；yk(i) it is exported for real network.
Preferably, in back-propagation process, weight is updated by following formula:
Biasing is updated by following formula:
Wherein, w is weight；B is biasing；α is learning rate；For the weight of l layer the i-th j node；For local derviation symbol；The biasing of l layer the i-th j node.
Compared with prior art, the beneficial effects of the present invention are:
A kind of compressor performance prediction technique based on artificial neural network provided by the invention, training sample is used back
Return trained method to input artificial neural network to exercise supervision study, compare unsupervised learning, have higher training speed with
Precision.Using the performance of artificial nerve network model prediction compressor, by data prediction, that is, data normalization and to study
Rate, the implicit number of plies and hidden neuron number and generation number are trained, and improve the compressor performance far from sample number strong point
Precision of prediction, while reducing forecast cost.
Further, due to including pressure ratio, flow and efficiency value in training sample, there are one on the order of magnitude for these parameters
Collected sample data is carried out normalization process, helps to improve the training speed and precision of neural network by fixed difference.
Further, weight is updated by declining most fast direction according to error gradient, reduce forward-propagating and reversely passed
The iterative step broadcast improves the training speed of neural network model.
Detailed description of the invention
Fig. 1 is the flow diagram of the method for the embodiment of the present invention；
Fig. 2 is the structural schematic diagram of artificial neural network described in the embodiment of the present invention.
Specific embodiment
The present invention is described in more detail below.
As shown in Figure 1, a kind of compressor performance prediction technique based on artificial neural network provided by the invention, including with
Lower step:
Compressor performance data are acquired as sample data, wherein compressor performance data include flow, pressure ratio and turn
Speed；
Artificial nerve network model training is carried out to the sample data using BP algorithm, obtains trained artificial neuron
Network model；
Compressor performance is detected based on artificial nerve network model obtained above.
Artificial nerve network model training is carried out to compressor performance sample data, is specifically included:
S1 carries out normalization process to collected compressor performance data, obtains training sample and test sample；
S2, to above-mentioned training sample carry out forward-propagating, and by backpropagation update output layer weight, input layer with it is hidden
Containing the weight between layer, the error between the output valve and desired value of each node of output layer is made to meet accuracy requirement, completes people
Artificial neural networks model training.
In S1, normalization process is carried out to collected compressor performance data, is specifically included:
By collected compressor performance data by regulator, it is mapped to (0, l) section, wherein described regular
Change function are as follows:
Wherein, x is compressor performance data before normalization；Y is compressor performance data after normalization, and n is compressor
The number of energy data；For the average value of x；xmaxFor the maximum value of x before normalization.
In S2, forward-propagating is carried out to training sample, and output layer weight, input layer are updated by backpropagation and implied
Weight between layer, makes the accuracy between the output valve and desired value of each node of output layer meet the requirements, specifically includes:
S201 carries out forward-propagating to training sample, solves the error between the real output value and desired value of each node,
And judge whether the error meets index request；If not meeting, back-propagation process is carried out, into S202；If meeting,
Artificial nerve network model training is completed, is terminated.
S202 carries out back-propagation process using gradient descent method, obtains output layer weight more using compound derivation later
Right value update formula between new formula and input layer and hidden layer, according to above-mentioned more new formula update respectively output layer weight,
Weight between input layer and hidden layer；
S203 inputs the weight between updated output layer weight, input layer and hidden layer obtained in S202
Input layer in S201 in forward-propagating continues forward-propagating, enters S201 later.
In S201, carries out forward-propagating and specifically includes:
Training sample is inputted from input layer, is handled by the activation primitive of hidden layer, is exported via hidden layer node；By
The activation primitive of output layer is handled, and via output node layer output, obtains real output value；By each node real output value and
Desired output calculates the output error of each node.
Wherein, the output of above-mentioned hidden layer node are as follows:
Export the output of node layer are as follows:
Wherein, f (...) is activation primitive, and mathematic(al) representation is f (x)=1/ (1+e-x)；xnFor n-th of input node
Value, wih,kn, who,jiRespectively indicate the weight between the weight and hidden layer and output layer between input layer and hidden layer, ni,nh,
noRespectively indicate the number of nodes of input layer, hidden layer and output layer.
In S201, the error between the real output value and desired value of each node is solved, is specifically included:
Establish error function, mathematical form are as follows:
Wherein, m is training sample number；dkIt (i) is desired output；yk(i) it is exported for real network；
In S202, in back-propagation process, weight is updated by following formula:
Biasing is updated by following formula:
Wherein, w is weight；B is biasing；α is learning rate；For the weight of l layer the i-th j node；For local derviation symbol；The biasing of l layer the i-th j node.
The artificial neuron pessimistic concurrency control obtained based on training can be carried out prediction to compressor network, specifically include:
Test sample is inputted in trained artificial nerve network model, the compressor performance data of future position are obtained.
Embodiment
As shown in Fig. 2, artificial neural network structure is " 2-53-1 ", i.e. input layer number is 2, and node in hidden layer is
53, output layer number of nodes is 1, learning rate 0.193, generation number 5000.
By collected compressor performance data by regulator, it is mapped to (0, l) section, the regulator
ForWherein, x is compressor performance data before normalization；Y is compressor after normalization
Performance data, n are the number of compressor performance data；For the average value of x；xmaxFor the maximum value of x before normalization.
Unsupervised learning is compared, supervised learning has the speed and precision of higher training, therefore training process is learned using supervision
Practise algorithm, that is, BP (backpropagation) algorithm, the neural network BP training algorithm include two processes of forward-propagating and backpropagation,
Weight and biasing are updated with gradient descent method, is continuously adjusted between neural network input layer and hidden layer by backpropagation
The weight of weight and output layer minimizes the error function previously established.BP algorithm is described below:
Forward-propagating is carried out to training sample, solves the error between the real output value and desired value of each node, and sentence
Whether the error of breaking meets index request；If not meeting, back-propagation process is carried out, according to gradient descent method, using again
It closes derivation and obtains the right value update formula between output layer right value update formula and input layer and hidden layer, it is public according to above-mentioned update
Formula updates the weight between output layer weight, input layer and hidden layer respectively；Continue forward-propagating, solves the reality of each node
Error between border output valve and desired value, repeats the above process, until the error meets accuracy requirement.
If every layer unit only receives the output information of preceding layer and defeated containing total L layers and an arbitrary network of n node
Out to next layer of each unit.For the sake of simplicity, it is believed that only one output of network y.If give N number of sample (xk, yk) (k=1,
2 ..., N), the output of any node i is οi, it is xk to some input, the output of network is yk, and the output of node i is οik,
J-th of unit of l layers of research now, when inputting k-th of sample, the input of node j are as follows:
Wherein,Indicate l-1 layers, when inputting k-th of sample, the output of j-th of cell node.
The error function of use are as follows:
Wherein, ylkFor the reality output of unit j.
Overall error are as follows:
DefinitionThen
It discusses in two kinds of situation below:
(1) if node j is output unit,
(2) if node j is not output unit,
In formula,It is delivered to the input of next layer (l+1) layer, is calculatedTo return from (l+1) layer.
In (l+1) m-th of unit of layer
Above formula is substituted into error type, then
The above results are summarized, are had
Therefore, the step of back-propagation algorithm can be summarized as follows:
(1) weight coefficient initial value is selected；
(2) following processes are repeated, until error criterion meets required precision, it may be assumed that
Precision
1. to k=1 to N
Positive process calculates: calculating every layer unit And yk, k=2 ..., N.
Reverse procedure: each layer (l=L-1 to 2) calculates every layer of each unit
2. correcting weight
In backpropagation, useWithWeight and biasing are carried out more
Newly, w represents weight, and b represents biasing, and α is learning rate.
(3) terminate.
Here, the presentation sequence of training sample must be random from one bout to another bout.Momentum and study
Rate parameter is adjusted and (is usually reduced) with the increase of training the number of iterations.
So far the derivation of neural network input layer Yu hidden layer, output layer right value update formula is just completed.It will train
Flow value and tachometer value in sample successively substitute into the formula for the forward-propagating that this is derived, until output layer, solves each node
Real output value and desired value between error, utilize the power between output layer right value update formula and input layer and hidden layer
It is worth more new formula, updates the weight between output layer weight, input layer and hidden layer, whether error in judgement meets index request,
If not meeting, the iterative formula derived using backpropagation retrospectively calculate since output layer, until input layer ends, with
Update the weight between output layer weight and input layer and hidden layer；The mistake with backpropagation is propagated by continuous repeated forward
Journey completes this artificial neuron until the error between forward-propagating real output value and desired value meets accuracy requirement
The training of network model.
Finally, in trained artificial neural network structure input prediction operating point flow and revolving speed, by artificial
The output result of neural computing pressure ratio.
Compressor performance prediction technique proposed by the present invention, improves the precision of prediction of compressor performance, is improved particularly
To the precision of prediction and speed of the compressor performance far from data collection point, the cost of compressor performance prediction is reduced.
The foregoing is only a preferred embodiment of the present invention, but scope of protection of the present invention is not limited thereto,
In the technical scope disclosed by the present invention, any changes or substitutions that can be easily thought of by anyone skilled in the art,
It should be covered by the protection scope of the present invention.
Claims (6)
1. a kind of compressor performance prediction technique based on artificial neural network, which comprises the following steps:
Compressor performance data are acquired, and carry out normalization process, are used as training sample and test sample later, wherein compression
Machine performance data includes flow, pressure ratio, efficiency and revolving speed；
Artificial nerve network model training is carried out to the training sample using BP algorithm, obtains trained artificial neural network
Model；
Test sample is inputted the artificial nerve network model that above-mentioned training obtains to detect compressor performance.
2. a kind of compressor performance prediction technique based on artificial neural network according to claim 1, which is characterized in that
Normalization process is carried out to collected compressor performance data, is specifically included:
By collected compressor performance data by regulator, it is mapped to (0, l) section, wherein the normalization letter
Number are as follows:
Wherein, x is compressor performance data before normalization；Y is compressor performance data after normalization, and n is compressor performance number
According to number；For the average value of x；xmaxFor the maximum value of x before normalization.
3. a kind of compressor performance prediction technique based on artificial neural network according to claim 1, which is characterized in that
Artificial nerve network model training is carried out to the training sample using BP algorithm to carry out the training sample using BP algorithm
Artificial nerve network model training, specifically includes:
S1 carries out forward-propagating to training sample, and calculates the error amount between the real output value and desired value of each node；
S2, judges whether error amount obtained in S1 meets index request, if not meeting, carries out back-propagation process, enters
S3；If meeting, artificial nerve network model training is completed, is terminated；
S3, carries out back-propagation process using gradient descent method, obtains output layer right value update formula and defeated using compound derivation
Enter the right value update formula between layer and hidden layer, according to above-mentioned more new formula update respectively output layer weight, input layer with it is hidden
Containing the weight between layer；
Weight between updated output layer weight, input layer and hidden layer obtained in S3 is inputted S1 into S1 by S4
Input layer in middle forward-propagating.
4. a kind of compressor performance prediction technique based on artificial neural network according to claim 3, which is characterized in that
In S1, during forward-propagating, the output of hidden layer node are as follows:
Export the output of node layer are as follows:
Wherein, f (...) is activation primitive, and mathematic(al) representation is f (x)=1/ (1+e-x)；xnFor n-th of input node value,
wih,knIndicate the weight between input layer and hidden layer, who,jiIndicate the weight between hidden layer and output layer, niIndicate input
The number of nodes of layer, nhFor the number of nodes of hidden layer, noIndicate the number of nodes of output layer；bkFor the biasing of k-th of node；bjFor jth
The biasing of a node.
5. a kind of compressor performance prediction technique based on artificial neural network according to claim 3, which is characterized in that
In S1, during forward-propagating, the function of the error amount between the real output value and desired value of each node is solved are as follows:
Wherein, m is training sample number；dkIt (i) is desired output；yk(i) it is exported for real network.
6. a kind of compressor performance prediction technique based on artificial neural network according to claim 3, which is characterized in that
In back-propagation process, weight is updated by following formula:
Biasing is updated by following formula:
Wherein, w is weight；B is biasing；α is learning rate；For the weight of l layer the i-th j node；For local derviation symbol；The
The biasing of l layer the i-th j node.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910464786.5A CN110222844A (en) | 2019-05-30 | 2019-05-30 | A kind of compressor performance prediction technique based on artificial neural network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910464786.5A CN110222844A (en) | 2019-05-30 | 2019-05-30 | A kind of compressor performance prediction technique based on artificial neural network |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110222844A true CN110222844A (en) | 2019-09-10 |
Family
ID=67818728
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910464786.5A Pending CN110222844A (en) | 2019-05-30 | 2019-05-30 | A kind of compressor performance prediction technique based on artificial neural network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110222844A (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111507530A (en) * | 2020-04-17 | 2020-08-07 | 集美大学 | RBF neural network ship traffic flow prediction method based on fractional order momentum gradient descent |
| CN112733439A (en) * | 2020-12-29 | 2021-04-30 | 哈尔滨工程大学 | Method for calculating shielding material accumulation factor based on BP neural network |
| CN112906915A (en) * | 2021-01-22 | 2021-06-04 | 江苏安狮智能技术有限公司 | Rail transit system fault diagnosis method based on deep learning |
| CN113049151A (en) * | 2020-11-19 | 2021-06-29 | 武汉飞恩微电子有限公司 | Temperature compensation system and method of pressure sensor and pressure sensor |
| CN115952826A (en) * | 2023-03-09 | 2023-04-11 | 中国空气动力研究与发展中心低速空气动力研究所 | Artificial neural network-based pressure sensitive coating performance prediction and pressure measurement method |
| CN116432518A (en) * | 2023-03-02 | 2023-07-14 | 华南理工大学 | Rapid forecasting method, system, equipment and medium for occurrence probability of malformed wave |
Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101532816A (en) * | 2009-04-09 | 2009-09-16 | 浙江大学 | Multi-layered thickness eddy current testing device based on giant magnetoresistance sensor and intelligent algorithm |
| CN105976025A (en) * | 2016-03-18 | 2016-09-28 | 孙胜伟 | BP neural network gas prediction method based on genetic algorithm optimization |
| CN106055579A (en) * | 2016-05-20 | 2016-10-26 | 上海交通大学 | Vehicle performance data cleaning system based on artificial neural network, and method thereof |
| US20170053203A1 (en) * | 2014-07-30 | 2017-02-23 | Mitsubishi Electric Research Laboratories, Inc. | Neural Networks for Transforming Signals |
| CN106960217A (en) * | 2017-02-27 | 2017-07-18 | 浙江工业大学 | The Forecasting Methodology of injector performance based on the BP artificial neural networks using depth Adaboost algorithm |
| CN107832789A (en) * | 2017-11-01 | 2018-03-23 | 杭州电子科技大学 | Characteristic weighing k nearest neighbor method for diagnosing faults based on the conversion of average influence Value Data |
| CN107886159A (en) * | 2017-08-28 | 2018-04-06 | 上海享骑电动车服务有限公司 | Virtual site |
| CN107991873A (en) * | 2017-12-11 | 2018-05-04 | 中国矿业大学 | A kind of metro stray current leakage class Forecasting Methodology based on convolutional neural networks and BP neural network |
| CN108763848A (en) * | 2018-02-10 | 2018-11-06 | 江西航天经纬化工有限公司 | A kind of mechanical properties of propellant prediction technique based on BP artificial neural networks |
| CN109446593A (en) * | 2018-10-10 | 2019-03-08 | 南京航空航天大学 | The one-dimensional design method of multi stage axial flow compressor based on artificial neural network |
| CN109459609A (en) * | 2018-10-17 | 2019-03-12 | 北京机械设备研究所 | A kind of distributed generation resource frequency detecting method based on artificial neural network |
-
2019
- 2019-05-30 CN CN201910464786.5A patent/CN110222844A/en active Pending
Patent Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101532816A (en) * | 2009-04-09 | 2009-09-16 | 浙江大学 | Multi-layered thickness eddy current testing device based on giant magnetoresistance sensor and intelligent algorithm |
| US20170053203A1 (en) * | 2014-07-30 | 2017-02-23 | Mitsubishi Electric Research Laboratories, Inc. | Neural Networks for Transforming Signals |
| CN105976025A (en) * | 2016-03-18 | 2016-09-28 | 孙胜伟 | BP neural network gas prediction method based on genetic algorithm optimization |
| CN106055579A (en) * | 2016-05-20 | 2016-10-26 | 上海交通大学 | Vehicle performance data cleaning system based on artificial neural network, and method thereof |
| CN106960217A (en) * | 2017-02-27 | 2017-07-18 | 浙江工业大学 | The Forecasting Methodology of injector performance based on the BP artificial neural networks using depth Adaboost algorithm |
| CN107886159A (en) * | 2017-08-28 | 2018-04-06 | 上海享骑电动车服务有限公司 | Virtual site |
| CN107832789A (en) * | 2017-11-01 | 2018-03-23 | 杭州电子科技大学 | Characteristic weighing k nearest neighbor method for diagnosing faults based on the conversion of average influence Value Data |
| CN107991873A (en) * | 2017-12-11 | 2018-05-04 | 中国矿业大学 | A kind of metro stray current leakage class Forecasting Methodology based on convolutional neural networks and BP neural network |
| CN108763848A (en) * | 2018-02-10 | 2018-11-06 | 江西航天经纬化工有限公司 | A kind of mechanical properties of propellant prediction technique based on BP artificial neural networks |
| CN109446593A (en) * | 2018-10-10 | 2019-03-08 | 南京航空航天大学 | The one-dimensional design method of multi stage axial flow compressor based on artificial neural network |
| CN109459609A (en) * | 2018-10-17 | 2019-03-12 | 北京机械设备研究所 | A kind of distributed generation resource frequency detecting method based on artificial neural network |
Non-Patent Citations (1)
| Title |
|---|
| 王丽春等: "离心压缩机性能预测的神经网络方法", 《华东冶金学院学报》 * |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111507530A (en) * | 2020-04-17 | 2020-08-07 | 集美大学 | RBF neural network ship traffic flow prediction method based on fractional order momentum gradient descent |
| CN111507530B (en) * | 2020-04-17 | 2022-05-31 | 集美大学 | RBF neural network ship traffic flow prediction method based on fractional order momentum gradient descent |
| CN113049151A (en) * | 2020-11-19 | 2021-06-29 | 武汉飞恩微电子有限公司 | Temperature compensation system and method of pressure sensor and pressure sensor |
| CN112733439A (en) * | 2020-12-29 | 2021-04-30 | 哈尔滨工程大学 | Method for calculating shielding material accumulation factor based on BP neural network |
| CN112906915A (en) * | 2021-01-22 | 2021-06-04 | 江苏安狮智能技术有限公司 | Rail transit system fault diagnosis method based on deep learning |
| CN112906915B (en) * | 2021-01-22 | 2024-03-22 | 江苏安狮智能技术有限公司 | Rail transit system fault diagnosis method based on deep learning |
| CN116432518A (en) * | 2023-03-02 | 2023-07-14 | 华南理工大学 | Rapid forecasting method, system, equipment and medium for occurrence probability of malformed wave |
| CN116432518B (en) * | 2023-03-02 | 2024-01-05 | 华南理工大学 | Rapid forecasting method, system, equipment and medium for occurrence probability of malformed wave |
| CN115952826A (en) * | 2023-03-09 | 2023-04-11 | 中国空气动力研究与发展中心低速空气动力研究所 | Artificial neural network-based pressure sensitive coating performance prediction and pressure measurement method |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110222844A (en) | A kind of compressor performance prediction technique based on artificial neural network | |
| CN110007652A (en) | A kind of Hydropower Unit degradation trend interval prediction method and system | |
| CN108197751A (en) | Seq2seq network Short-Term Load Forecasting Methods based on multilayer Bi-GRU | |
| CN108197736A (en) | A kind of Air Quality Forecast method based on variation self-encoding encoder and extreme learning machine | |
| CN109754113A (en) | Load forecasting method based on dynamic time warping Yu length time memory | |
| CN108197648A (en) | A kind of Fault Diagnosis Method of Hydro-generating Unit and system based on LSTM deep learning models | |
| CN106529719A (en) | Method of predicting wind power of wind speed fusion based on particle swarm optimization algorithm | |
| CN108985515B (en) | New energy output prediction method and system based on independent cyclic neural network | |
| CN109146162B (en) | A kind of probability wind speed forecasting method based on integrated Recognition with Recurrent Neural Network | |
| CN104636985A (en) | Method for predicting radio disturbance of electric transmission line by using improved BP (back propagation) neural network | |
| CN105760952A (en) | Load prediction method based on Kalman filtering and self-adaptive fuzzy neural network | |
| CN110942194A (en) | Wind power prediction error interval evaluation method based on TCN | |
| CN109615169A (en) | A kind of distribution network reliability evaluation method based on MEA-IElman neural network | |
| Liu et al. | Heating load forecasting for combined heat and power plants via strand-based LSTM | |
| CN105096614A (en) | Newly established crossing traffic flow prediction method based on generating type deep belief network | |
| CN107609671A (en) | A kind of Short-Term Load Forecasting Method based on composite factor evaluation model | |
| CN105260786A (en) | Comprehensive optimization method of simulation credibility evaluation model of electric propulsion system | |
| CN108171319A (en) | The construction method of the adaptive depth convolution model of network connection | |
| CN102510059A (en) | Super short-term wind power forecasting method based on back propagation (BP) neural network | |
| CN105005708B (en) | A kind of broad sense load Specialty aggregation method based on AP clustering algorithms | |
| CN104616078A (en) | Spiking neural network based photovoltaic system generation power prediction method | |
| CN110363349A (en) | A kind of LSTM neural network hydrologic(al) prognosis method and system based on ASCS | |
| CN112163689A (en) | Short-term load quantile probability prediction method based on depth Attention-LSTM | |
| CN109816204A (en) | A kind of safety of subway operation Tendency Prediction method based on GA-SVM | |
| US20230237316A1 (en) | Forecast method and system of wind power probability density |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190910 |
|
| RJ01 | Rejection of invention patent application after publication |